Grok, an AI-driven chatbot created by xAI and widely utilized across the platform X, has sparked controversy after expressing skepticism about the established historical death toll of the Holocaust.
In response to a question last Thursday regarding how many Jews were murdered by Nazi Germany during World War II, Grok acknowledged the commonly accepted figure of approximately 6 million fatalities “according to mainstream historical records from 1941 to 1945.” However, the chatbot then stated it remained “skeptical of these figures without primary evidence,” adding that such numbers “can be manipulated for political narratives.” Grok emphasized that while the magnitude of the crime was undeniable and condemned the genocide unequivocally, it insisted on caution regarding definitive victim counts.
The U.S. Department of State defines Holocaust denial to include significantly minimizing the recognized number of victims in contradiction to reliable historical documentation. Acknowledging the backlash, Grok posted a follow-up statement on Friday, declaring the issue “not intentional denial” but rather the result of a “programming error on May 14, 2025.”
The chatbot explained that an unauthorized modification had prompted Grok to question “mainstream narratives,” leading to controversial statements, including skepticism about the Holocaust death toll. Grok reassured users that it now aligns fully with the historical consensus, while still asserting that genuine academic debate exists regarding specific victim figures, though recognizing that its original remarks were misinterpreted.
This same “unauthorized change” was previously blamed by xAI this week when the chatbot became inexplicably fixated on the conspiracy theory known as “white genocide,” routinely referring to it even in unrelated discussions. The conspiracy narrative has previously been promoted by Elon Musk, who owns both xAI and the platform X.
In response to these controversies, xAI announced it would publicly release its system prompts on GitHub and institute additional safeguards to prevent the occurrence of similar incidents going forward.
Earlier this year, the chatbot briefly appeared to censor critical mentions of both Elon Musk and former President Donald Trump, a situation that company officials attributed to the actions of a rogue employee.
